The role of relatives in RP is less clear. Relatives of people with bipolar disorder experience high levels of burden which are associated with physical and mental health problems and increased use of medical and mental health services [4], particularly amongst caregivers living with patients [5]. Among people with bipolar disorder, there is a perception that carers and families are often excluded from management decisions and ignored by health professionals to the distress of family members who remain uninformed about bipolar disorder [6]. Most families report wishing for support and education from services, but that they rarely receive it [7]. Under these circumstances, families cannot be expected to be as effective as they might be in detecting clinical signs of illness and obtaining help. There are several mechanisms through which relatives' involvement can support service users. Relatives can impact positively on the outcome for patients by providing structures that encourage stable routines and emotional self-regulation strategies [3]. Conversely, relatives' expressed emotion is a robust predictor of relapse in psychiatric conditions, particularly mood disorders [8]. High expressed emotion has been associated with dysfunctional patterns of communication [9] and blaming attributions for negative patient-related events [10].

Together these findings have prompted a growing field of research into interventions at the level of the family to reduce carer burden, develop more helpful illness attributions and patterns of family communication, improve medication concordance and reduce relapse rates. A systematic review of interventions involving relatives was unable to draw conclusions due the heterogeneity and limited size of trials [11]. Nevertheless, recent trials conducted in families of adults [12] and adolescents [13] or carers alone [14] have yielded positive effects on outcome, illustrating the potential value of involving relatives to improve the outcome of bipolar disorder. It has been recommended that engaging families in helping patients to recognize individual early warning signs of mania or depression is a helpful adjunct to pharmacological management [1,3].

Typically, however, research into relapse prevention interventions for bipolar disorder has not specifically sought to involve relatives or carers, but has assessed individualized treatment delivered through specialist services, expert therapists or extensive therapy [15-17] none of which are routinely available in the mainstream services such as the UK National Health Service (NHS). Moreover, RP planning is most useful when patients are well, which is a time when they are likely to have limited contact with medical or mental health specialists. During these periods, service users' primary contact will be a designated member of their community mental health team, who is responsible for their case management. These care coordinators are typically from a nursing, occupational therapy or social work background and will have limited opportunities for specialist training in specific psychological interventions for bipolar disorder [18]. This model is typical within the UK NHS for community follow-up care for people with serious mental illness, and is increasingly found in many over services across the world [19].

A key advantage of RP is that, compared to more sophisticated approaches involving early warning signs (such as some forms of cognitive behaviour therapy and family therapy) simple RP interventions can be taught more quickly and easily to both non-specialist health professionals without requiring extensive training in psychotherapy [20]. A recent trial found that RP could be taught to care coordinators and that this improved social functioning compared with treatment as usual amongst service users with bipolar disorder [21].

Consequently opportunities to involve relatives in relapse prevention planning are likely to most usefully involve care coordinators, who are not trained in family therapy and may not recognize the potential benefit of engaging family members in patients' care planning. Attempts to involve relatives in relapse planning have however been met with limited success [21]. If the potential benefits of involving relatives in RP is to be achieved within routine care, it is important to understand the value health professionals, patients and relatives see (if any) in involving family members in relapse prevention planning, and what barriers exist that deter relatives from taking a greater role.

The quality of the relationships between relatives and service users had the potential to simplify or increase the complexity of delivering relapse prevention. It is known that families of individuals with bipolar disorder often experience high levels of family burden and criticism, similar to families of individuals with other long term mental health problems [5] and often feel isolated in attempting to balance their own needs with those of their family members' and unsupported by professionals [27]. Our findings suggest even with further trial evidence of the effectiveness of involving families in relapse prevention, this may not be appropriate (or possible) until the needs of carers and relatives are also addressed. Moreover, that health care professionals should receive appropriate training and support to provide this. To engage family members, health professionals may firstly need to establish independent relationships with family members before attempting to engage them in systemic interventions. Relatives may first require individual support to address their own mental health needs. Interventions may also be needed that aim to improve communication and day to day interactions within the family. A clear rationale for family involvement and the specific role of the relative is also crucial. Research has demonstrated that whilst family members desire more involvement and education [7] they can feel that health professionals' focus on encouraging service user autonomy renders them powerless and without a role [28]. Our study demonstrates that preserving autonomy and privacy is bidirectional and similarly important for service users and relatives and is another potential barrier to involving relatives in relapse prevention planning. These tasks may be complex and require therapists with skills, time and appropriate supervision. In our study, care coordinators were able to see the value of involving relatives in terms of saving input from services in the longer term; however they were concerned about the feasibility of increasing the frequency and duration of visits, and the need to take on relatives as an invisible caseload. Without the support of managers, involving relatives in care in this way is unlikely to be implemented into practice. Further research is needed as to the value placed on relapse prevention (and potential barriers) at the level of the wider team organization [29].

A qualitative approach was employed which provided rich data grounded in stakeholders views and ideas [30]. With 52 participants, the sample was large and thematic saturation was assured. A particular strength of the investigation was the use of triangulation which is a recognized procedure for increasing trustworthiness of qualitative analysis [25]. Triangulation was sought in two ways. Firstly, the study question was approached from the perspectives of service providers, users and relatives. Secondly, the thematic framework was developed and tested by a multidisciplinary team of researchers [30].